How to conditionally create a new batch using an existing batch in Catalog to Annotate

Python SDK
,
1

Here’s my situation:

  • I create a list of assets containing Text data rows (let’s say n data rows).
  • I upload these assets to Catalog as a batch called B1
  • I annotate all data rows in B1
  • I export the annotated data rows into my code on my machine and do some operations based on the labels and some operations that can’t be done on Labelbox (so we’re not talking about simple filtering). And let’s say, out of the n initial data rows I had, I decided to leave out m data rows.
  • Now, I want to do a new round of annotation with new annotators and new schema on those n-m data rows (let’s call this new batch, B2). But I don’t want to upload new assets to the Catalog, instead, I want to create a new batch which is a sample of B1 but only with the data rows I kept after my internal operations (reason? because I don’t want to get charged new LBUs and use data rows that are already uploaded, labeled, and for which I got charged.)
  • The question is how can I achieve this? As far as I know, If I decide to upload new data rows that are a subset of an existing batch I will get a global key error. And if I create new global keys then I will get charged again.
2

Hi @pedram,

You can create a new project and send the batch of data rows to the new project. When you send the batch to the new project, make sure to exclude current annotations. You won’t be charged for LBUs in Catalog but there will be charges for Annotate.

Best,
Janny

3

Thanks for your response. So even if I don’t get charged for uploading new data to Catalog if I use data rows with existing global keys, I will still be charged if I send the same data to a new project to get more annotation? Is that right?

4

Hi @pedram ,

You will only be charged for each data rows that you label within the project. More information here! Manage account and billing.

Best,
Janny